Abstracts

Enabling data-driven research and development in composite product engineering : An over-braiding case study

by Daniel Griffin

The fibre reinforced polymer (FRP) composites manufacturing industry stands to benefit significantly from the exploitation of data-driven research and development practices. However, research organisations which are best placed to make advances in the associated engineering disciplines are often ill-equipped to generate and utilise the prerequisite consistent, contextualised data sets. The work herein seeks to address this issue by demonstrating a progressive approach to the generation of manufacturing data sets, in which emphasis is placed on maximising the potential research value, i.e. usefulness, of those data. The approach is demonstrated via the development and implementation of a bespoke in-process measurement capability for the 2D axial over-braiding preforming process, with which to automate the measurement braid angle and preform thickness as informative quality metrics. Measurements were achieved using 2D Fourier transform image analysis and laser profilometry, respectively. System design prioritised contextualisation of measurement data and ensuring simplicity of operation. A method with which to evaluate the potential research value of additional process variable data sets, based on assessment of variable inter-dependencies, was developed to support the design process. Through it, a number of process parameters were identified that represent valuable, yet often overlooked, contextual information regarding process quality; the functionality to record these metadata via the user interface was thus included. Measurement system sensitivity to a range of material, process and procedural factors was assessed in order to optimise measurement performance and evaluate the associated uncertainty. Insights generated into the effect of yarn twist on preform thickness, factors influencing the stabilisation of the deposited braid architecture, and the relationship between braid angle and convergence cone geometry are presented. These illustrate some of the benefits that are enabled by ensuring the availability and re-useability of consistent data sets, as required to perform of data-driven research and development.